Research map / Probabilistic and scientific modeling / Bayesian optimization
Acquisition function design
Develop new acquisition criteria and stopping rules for sequential Bayesian optimization.
97 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Bayesian optimization, keyword cluster “acquisition · bayesian · multi objective”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
26 papers in 2023–24 and 29 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 28.3% to 21.6%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Bayesian Optimization with Robust Bayesian Neural Networks (NeurIPS 2016 · 263 citations)
- 2017 · Practical Bayesian Optimization for Model Fitting with Bayesian Adaptive Direct Search (NeurIPS 2017 · 157 citations)
- 2018 · Stagewise Safe Bayesian Optimization with Gaussian Processes (ICML 2018 · 65 citations)
- 2019 · Practical Two-Step Lookahead Bayesian Optimization (NeurIPS 2019 · 20 citations)
- 2020 · Multi-Fidelity Bayesian Optimization via Deep Neural Networks (NeurIPS 2020 · 22 citations)
- 2021 · Mercer Features for Efficient Combinatorial Bayesian Optimization (AAAI 2021 · 16 citations)
- 2022 · Monte Carlo Tree Search based Variable Selection for High Dimensional Bayesian Optimization (NeurIPS 2022 · 10 citations)
- 2023 · Unexpected Improvements to Expected Improvement for Bayesian Optimization (NeurIPS 2023 · 48 citations)
- 2024 · A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · 5 citations)
- 2025 · BOIDS: High-Dimensional Bayesian Optimization via Incumbent-Guided Direction Lines and Subspace Embeddings (AAAI 2025 · 2 citations)
- 2026 · Goal-Oriented Lower-Tail Calibration of Gaussian Processes for Bayesian Optimization (ICML 2026 · 0 citations)
Earliest papers here
- Bayesian Optimization with Robust Bayesian Neural Networks (NeurIPS 2016 · 263 citations)
- Bayesian optimization under mixed constraints with a slack-variable augmented Lagrangian (NeurIPS 2016 · 49 citations)
- Batched Gaussian Process Bandit Optimization via Determinantal Point Processes (NeurIPS 2016 · 32 citations)
What has been tried
77 papers here report 739 measured changes (a component added, removed or swapped, compared with a baseline): helped 61%, no clear effect 10%, hurt 22%, mixed 7%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Use a problem-tailored HMC network architecture rather than the default architecture | Pest Control · average maximum value | Helped | A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · 5 citations) | |
| Use Thompson Sampling rather than UCB to approximate the maximizer set | Asynchronous synthetic benchmarks · predictive regret | Helped | Transition Constrained Bayesian Optimization via Markov Decision Processes (NeurIPS 2024 · 3 citations) | |
| add incumbent-guided direction lines instead of uniformly random guiding lines | LassoDNA-180D · optimization performance | Helped | BOIDS: High-Dimensional Bayesian Optimization via Incumbent-Guided Direction Lines and Subspace Embeddings (AAAI 2025 · 2 citations) | |
| Use qREI only during restarts, retaining standard initialization | HPA101-2 and MOPTA08 · objective value | No clear effect | Regional Expected Improvement for Efficient Trust Region Selection in High-Dimensional Bayesian Optimization (AAAI 2025 · 2 citations) | |
| Initialize length-scales with ℓ₀ = √d (c=1) instead of ℓ₀=0.693 | High-dimensional optimization benchmarks · optimization performance | Helped | Standard Gaussian Process is All You Need for High-Dimensional Bayesian Optimization (ICLR 2025 · 2 citations) | |
| Use ColaBO with a Matérn-5/2 kernel instead of a squared-exponential kernel | Synthetic functions · aggregate optimization performance | Hurt | A General Framework for User-Guided Bayesian Optimization (ICLR 2024 · 2 citations) | |
| Replace deep-ensemble constraint classification with a GP classifier | Half of the benchmark problems · best objective value | Hurt | Boundary Exploration for Bayesian Optimization With Unknown Physical Constraints (ICML 2024 · 2 citations) | |
| Use GBDT-native difficulty estimation instead of ERC | Eight DBMS benchmarks · relative performance gain vs SMAC | 6.6% | Helped | CoffeeBoost: Gradient Boosting Native Conformal Inference for Bayesian Optimization (AAAI 2025 · 1 citation) |
| use structure-aware hyperparameter embeddings instead of values alone | SVM and XGBoost tasks, average ranking over 5 datasets · ranking | Helped | Modeling All Response Surfaces in One for Conditional Search Spaces (AAAI 2025 · 1 citation) | |
| Increase GP-UCB initial training set from 100 to 500 points | GB1(4) · best-so-far log fitness | Helped | Optimistic Games for Combinatorial Bayesian Optimization with Application to Protein Design (ICLR 2025 · 1 citation) | |
| add Thompson sampling for exploration | Forrester and quadratic synthetic functions · optimization performance | Helped | MALIBO: Meta-learning for Likelihood-free Bayesian Optimization (ICML 2024 · 1 citation) | |
| Use antithetic coupling in the MLMC estimator | 1D toy example · computational complexity | O(ε^{-2}(log ε)^2) -> O(ε^{-2}) | Helped | Accelerating Look-ahead in Bayesian Optimization: Multilevel Monte Carlo is All you Need (ICML 2024 · 1 citation) |
What reviewers will expect
Datasets and baselines named in the 21 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Branin · 4 of 21 papers
- Hartmann · 4 of 21 papers
- Ackley · 3 of 21 papers
- Rosenbrock · 3 of 21 papers
- MOPTA08 · 2 of 21 papers
- MNIST · 2 of 21 papers
- DTLZ2 · 2 of 21 papers
- Sushi preference dataset · 2 of 21 papers
Baselines
- Random · 4 of 21 papers
- HESBO · 3 of 21 papers
- SAASBO · 3 of 21 papers
- TuRBO · 3 of 21 papers
- CMA-ES · 3 of 21 papers
- ALEBO · 2 of 21 papers
- RDUCB · 2 of 21 papers
- SMAC · 2 of 21 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 55 papers; from OpenAlex and the papers' affiliations).
- Meta · 6 papers
- Google DeepMind · 4 papers
- Swiss Federal Institute of Technology · 3 papers
- ETH Zurich · 3 papers
- Albert-Ludwigs-Universität Freiburg · 3 papers
- California Institute of Technology · 3 papers
- Huawei · 2 papers
- The Chinese University of Hong Kong · 2 papers
- National University of Singapore · 2 papers
- Massachusetts Institute of Technology · 2 papers
Use it
Download BibTeX of 40 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Local and high-dimensional Bayesian optimization (90 papers)
- Multi-fidelity hyperparameter optimization (44 papers)
- Multi-objective Bayesian optimization (33 papers)
- Regret bounds for Bayesian optimization (43 papers)
Elsewhere in the topic
- Offline black-box optimization with surrogates (56 papers)
- High-dimensional Bayesian optimization (34 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).